vllm-backport-cherry-pick

name	vllm-backport-cherry-pick
description	Auto cherry-pick backport candidates and create a draft PR on the downstream repo. Use after scoring to attempt clean cherry-picks for ai-fixable candidates. The agent must still do semantic validation on the result.
compatibility	Requires git, gh CLI (authenticated), python3, bash
allowed-tools	Bash Read
metadata	{"author":"ZhanqiuHu","version":"1.0","tags":"vllm, backport, triage"}

Cherry Pick

Selects eligible candidates from ranked.json, attempts cherry-pick on each, and creates a draft PR if any succeed.

Candidate Selection

A PR is eligible if ALL of:

• backport_ease == "ai-fixable"
• score >= 50
• already_backported == false
• verdict is must_backport or likely_relevant

Usage

bash scripts/cherry-pick.sh \
  --input artifacts/backport-triage/ranked.json \
  --downstream /path/to/downstream-repo \
  --branch rhai/0.13.0 \
  --jira-url "https://redhat.atlassian.net/browse/..." \
  --report-url "https://github.com/..." \
  --output artifacts/backport-triage/cherry-pick-result.json

Output

cherry-pick-result.json:

{
  "status": "created|skipped",
  "pr_url": "https://...",
  "succeeded": 3,
  "conflicts": 1,
  "results": [{"number": 12345, "title": "...", "score": 85, "status": "success|conflict"}]
}

Agent Follow-up (Required)

After this skill runs, the agent MUST:

1. Semantic validation — review the cherry-picked diff, check imports reference modules that exist at the target tag, check for calls to post-release functions
2. If issues found, add a comment on the draft PR
3. Update the Jira ticket with the PR link
4. For conflict candidates with score >= 70, add label ai-autofix-candidate